japanese_tools/lib/iconv/codecs.dart

// https://en.wikipedia.org/wiki/JIS_encoding
// https://en.wikipedia.org/wiki/Japanese_Industrial_Standards

import 'dart:convert';
import 'dart:io';
import 'dart:typed_data';

import 'datatypes.dart';
import 'character_set_conversion.dart';

/// ISO-2002-JP (Usually abbreviated "JIS")
/// Encodes, decodes: JIS X 0202
///
///
// class ISO2002JP extends Encoding {}

enum ShiftJISCharacterSet {
  jisx0201,
  jisx0208,
  lundeRange,
}

/// Shift JIS (Usually abbreviated "SJIS")
/// Encodes, decodes: JIS X 0208
///
/// https://en.wikipedia.org/wiki/Shift_JIS
class ShiftJIS extends Encoding {
  int? decodeWithJsix0201(int char) {
    throw UnimplementedError();
  }

  /// Decode 16 bit character
  int? decodeWithJsix0208(int char) {
    int s1 = char >> 8;
    int s2 = char & 0x00FF;

    if (!((s1.between(0x81, 0x9f) || s1.between(0xe0, 0xea)) &&
        (s1.between(0x40, 0x7e) || s1.between(0x80, 0xfc)))) {
      return null;
    }

    int t1 = (s1 < 0xe0 ? s1 - 0x81 : s1 - 0xc1);
    int t2 = (s2 < 0x80 ? s2 - 0x40 : s2 - 0x41);

    int ffoo = 2 * t1 + (t2 < 0x5e ? 0 : 1) + 0x21;
    int ooff = (t2 < 0x5e ? t2 : t2 - 0x5e) + 0x21;
    int ffff = (ffoo << 8) + ooff;

    return JSIX0208Decoder().convert(ffff);
  }

  int? decodeWithLundeRange(int char) {
    throw UnimplementedError();
  }

  int? encodeWithJisx0201(int char) {
    throw UnimplementedError();
  }

  int? encodeWithJisx0208(int char) {
    final jsixEncoded = JSIX0208Encoder().convert(char);

    if (jsixEncoded == null) {
      return null;
    }

    int c1 = jsixEncoded >> 8;
    int c2 = jsixEncoded & 0x00FF;

    if (!(c1.between(0x21, 0x74) && c2.between(0x21, 0x7e))) {
      return null;
    }

    int t1 = (c1 - 0x21) >> 1;
    int t2 = ((((c1 - 0x21) & 1) != 0) ? 0x5e : 0) + (c2 - 0x21);

    int ffoo = (t1 < 0x1f ? t1 + 0x81 : t1 + 0xc1);
    int ooff = (t2 < 0x3f ? t2 + 0x40 : t2 + 0x41);

    return (ffoo << 8) + ooff;
  }

  int? encodeWithLundeRange(int char) {
    throw UnimplementedError();
  }

  @override
  List<int> encode(
    String input, {
    ShiftJISCharacterSet? forceCharacterSet,
    bool ignoreUnknownCharacters = false,
  }) {
    List<int> result = [];
    List<int> toEncode = input.codeUnits;
    for (int i = 0; i < toEncode.length; i++) {
      final int charcode = toEncode[i];
      final int? encodedChar = encodeWithJisx0208(charcode);
      if (encodedChar == null) {
        if (ignoreUnknownCharacters) {
          continue;
        } else {
          throw ConverterException(
            "Unable to convert character '${String.fromCharCode(charcode)}' at index [$i]",
          );
        }
      }
      result.add(encodedChar);
    }
    return result;
  }

  @override
  String decode(
    List<int> encoded, {
    ShiftJISCharacterSet? forceCharacterSet,
    bool ignoreUnknownCharacters = false,
  }) {
    List<int> result = [];
    for (int i = 0; i < encoded.length; i++) {
      final int encodedChar = encoded[i];
      final int? translatedChar = decodeWithJsix0208(encodedChar);
      if (translatedChar == null) {
        if (ignoreUnknownCharacters) {
          continue;
        } else {
          throw ConverterException(
            "Unknown character: ${encodedChar.hex} at index [$i]",
          );
        }
      }
      result.add(translatedChar);
    }
    return String.fromCharCodes(result);
  }

  @override
  Converter<String, List<int>> get encoder => throw UnimplementedError();

  @override
  Converter<List<int>, String> get decoder => throw UnimplementedError();

  @override
  String get name => "Shift JIS";
}

/// EUC-JP (Usually abbreviated "UJIS")
/// Unixized JIS
/// Encodes, decodes: JIS X 0201, JIS X 0208, JIS X 0212
///
/// https://en.wikipedia.org/wiki/Extended_Unix_Code#EUC-JP=
class EUCJP {}

void main() {
  final file = File("test.sjis.txt");
  final ByteData data = ByteData.sublistView(file.readAsBytesSync());
  final List<int> characterCodes = [
    for (int i = 0; i < ((data.lengthInBytes) / 2).floor(); i++)
      data.getUint16(i * 2)
  ];
  print(characterCodes);

  // print(ShiftJIS().decode(characterCodes));

  // print(JSIX0208Encoder().encode16BitCharacter('あ'.codeUnitAt(0)));
  // print(JSIX0208Encoder().encode16BitCharacter('幹'.codeUnitAt(0)));
  final x = ShiftJIS().encode("この文章は読めるかな？");
  print(data.buffer.asUint8List());
  File("test.sjis2.txt").writeAsBytesSync(Uint16List.fromList(x));
}